uzabase/LLM2Vec-Llama-2-7b-hf-mntp-unsup-simcse

Model Info

This is a model that applies LLM2Vec to Llama-2. Only the PEFT Adapter is distributed. LLM2Vec is fine-tuned on two tasks: MNTP and SimCSE, and this repository contains the results of applying SimCSE after MNTP. For the MNTP Adapter, please refer to this link.

Model Details

Model Description

• Model type: PEFT
• Language(s) (NLP): English
• License: Apache2.0
• Finetuned from model: Llama-2-7b-hf

Model Sources [optional]

• Repository: https://github.com/McGill-NLP/llm2vec
• Paper: https://arxiv.org/abs/2404.05961

Usage

• Please see original LLM2Vec repo

BenchMark

• Followings are summaries. Details are here

MTEB(Japansese)

| | Classification | Clustering | PairClassification | Reranking | BitextMining | Retrieval | Sts | AVG | | --- | ---| ---| ---| ---| ---| ---| ---| ---| ---| ---| ---| ---| ---| ---| | Llama2-Llm2vec-eng (This repo) | 0.527 | 0.258 | 0.501 | 0.217 | 0.275 | 0.296 | 0.765 | 0.408 | | Llama2-Llm2vec-jpn | 0.570 | 0.365 | 0.510 | 0.349 | 0.470 | 0.417 | 0.795 | 0.498 | | Swallow-Llm2vec-jpn | 0.621 | 0.391 | 0.510 | 0.475 | 0.475 | 0.491 | 0.832 | 0.523 |

MTEB(English)

	Classification	Clustering	Pair_Classification	Reranking	Retrieval	STS	平均
Llama2-Llm2vec-eng (this repo)	0.709	0.386	0.780	0.588	0.329	0.723	0.586
Llama2-Llm2vec-jpn	0.722	0.428	0.785	0.594	0.371	0.717	0.603
Swallow-Llm2vec-jpn	0.695	0.385	0.751	0.576	0.318	0.710	0.572

Training Details

Training Data

• Corpus for SimCSE from Wikipedia

Training Hyperparameter

• simcse_dropout: 0.3
• bidirectional: true
• pooling_mode: "mean"
• remove_unused_columns: false
• learning_rate: 3e-5
• loss_scale: 20
• batch_size: 256
• gradient_accumulation_steps: 1
• max_seq_length: 128
• lora_r: 16
• torch_dtype: "bfloat16"
• attn_implementation: "flash_attention_2"
• seed: 42
• bf16: true
• gradient_checkpointing: true

Accelerator Settings

• deepspeed_config:
  • gradient_accumulation_steps: 1
  • gradient_clipping: 1.0
  • offload_optimizer_device: nvme
  • offload_optimizer_nvme_path: /nvme
  • zero3_save_16bit_model: true
  • zero_stage: 2
• distributed_type: DEEPSPEED
• downcast_bf16: 'no'
• dynamo_config:
  • dynamo_backend: INDUCTOR
  • dynamo_mode: default
  • dynamo_use_dynamic: true
  • dynamo_use_fullgraph: true
• enable_cpu_affinity: false
• machine_rank: 0
• main_training_function: main
• mixed_precision: bf16
• num_machines: 1
• num_processes: 2
• rdzv_backend: static
• same_network: true
• quse_cpu: false

Framework versions

• Python: 3.12.3
• PEFT 0.11.1
• Sentence Transformers: 3.0.1
• Transformers: 4.41.0
• PyTorch: 2.3.0
• Accelerate: 0.30.1
• Datasets: 2.20.0
• Tokenizers: 0.19.1
• MTEB: 1.13.0